can you make an inference whitout having made any observations?

Answer 1

The answer depends on your definition of inference. In deductive reasoning observations are irrelevant or meaningless, since inferences follow naturally (or not) from given premises.
In inductive reasoning, inference is defined only with reference to a reliable generalization of a set of observations, in which case an inference without observation becomes meaningless. So, we see that in the former case inference and observation are mutually exclusive, in the latter mutually defined.

You may want to stop reading at this point... I know I will!

In the case of automatic logical inference, or so-called expert systems, the situation can become somewhat unclear. An expert system is a computer program that makes inferences based on prior knowledge (a knowledge base provided by human "experts"). The expert system can extend its knowledge base through input of new data from the real world. For example, an expert system at a nuclear reactor might infer that a core meltdown is imminent based on certain temperature and neutron flux measurements around the core. The inference is based on observation combined with preset threshhold values provided by experts.
Imagine next that a large computer is used to simulate the expert control system of the reactor. The programmers tell the simulation that a certain set of temperature and flux measurements have occured, and the simulated expert system makes an inference based on the knowledge base (the same base that was used by the real expert control system). The simulation has made an inference using virtual or simulated observations. Here the former (the actual real-world system) is of the inductive type, dependant by definition and design on observation. The latter is of the deductive type, depending only on the given premises, since no actual observations were made (and would be irrelevant anyway). However, even the real-world expert system could be viewed as hybrid deductive-inductive system, inasmuch as it requires both obervation and a certain set of premises. In this sense the system is not very dissimilar from economics or epidemiology, disciplines that depend heavily on statistics. And while the statistical vagaries of economies and plagues may give public officials and planners fits, it is entirely unnacceptable to be informed by a computer that the reactor may or may not be about to meltdown. Unfortunately, the realities of nuclear physics dictate that at least one possible set of circumstances could occur in which the expert control system would be unable to make a meaningful inference based soley on the threshhold values and observations. Here is an observation without an inference!
Now, we simulate this situation on yet a third computer, a meta-simulation. It uses real-world data (from reactor data archives) and the aforementioned simulation (yes- a simulation within a simulation!) to reach its conclusions. It concludes that there is at least one possible situation (which as far as we know has not yet occured) in which it cannot make a meaningful inference. This is Goedel's Incompleteness and Baudrillard's Simulacra attacking science and technology with a vengeance. Having simulated a simulation that discovers heretofor unknown reactor states, have we made inferences based on expert knowledge or on observation, or both? Or are the observations, since they are "virtual observations" never actually observed, "hyperreal" or simulacra-type observations, not true observations and therefore the inferences were made without observation? Yet the meta-simulation simulated a simulation based on real-world data, an observation-dependant system... now we have a viscious circle with no resolution. Goedel and Baudrillard would laugh at the insanity as the reactor melts, our simulation able only to tell us that it (the meltdown) had some undefinable statistical probability of occurence.
Of course, Chomsky might argue (along with the cognitive psychologists) that the "context-free grammar" of the various simulations makes unlikely any error in processing save that made by the human computer (i.e., brain). And, naturally, Vygotsky would tell us that any inference made by the simulation not shared with others (whether computer, simulation, or human) is not a real inference, since all knowledge (and therefore inferences, since inferences represent potentially new knowledge) only becomes real knowledge if it is socially mediated.
At last the answer! Inferences made by virtual observation (=non-observation) are not true inferences as long as they are not socially mediated (=shared), so their is no conflict... small consolation for the bath of radiation in which we sit as we scratch our heads trying to figure out what went wrong with the non-socially-mediating "expert control system"!

Answer 2

Is it about science or life/world in a big picture ?
You bet , we can . Human mind is possible of doing anything, not necessarily the right thing .
When inferences are made without having made any observations then its like a gamble . One can miss out on important details, the true facts . So, the outcome may or may not be acceptable . Its a 50-50 true/false situation .
Don't you ever find people around you doing that already ? I know.........every time its not possible to spend time on observation or getting first hand info.........well don't make any judgment then .

Answer 3

Yes. But that would an un-scientific method. In a scientific method you make observations and based on observation you will have to draw inference.

Answer 4

No. If you made no observations whatsoever how would you know what was going on at all? It could be anything in the whole world.

Answer 5

Yes, if someone else makes the observations.

Answer 6

Sports betting systems are sets of events that when combined for a particular game for a particular sport represents a profitable betting scenario

Answer 7

Yes. If you smell a garden of flowers and do not see it, then you are infering that a garden of flowers is within your space.

Answer 8

You can, but you risk having a better chance of being wrong.

Answer 9

yes, but will it be right ?